PeterThe filesystem used in the vast majority of Linux (and to a lesser extent on other Unix) distributions is the ext3 filesystem, which stands for extended filesystem.
Slashdot recently pointed to an article on IBM DeveloperWorks discussing the new ext4 filesystem currently in development, and some of its new features.
So what is a filesystem really? And why is the development of ext4 and the new features it brings to the table important for open source platforms?
Filesystems
A filesystem is a very important piece of software in your operating system. In essence, a filesystem is a method of organising files and directories on your hard disk and it allows you to actually store persistent data.
Modern filesystems allow you to do much more than just define a tree of directories and files and store data in them, though. Features such as access control and journalling (keeping a log of all events on the volume to make recovery easier) are some of the things that a filesystem does.
What's all the fuss about ext4?
Ext3 is probably the most popular filesystem in use on Linux distributions today. The reasons for that are that it has a lot of modern features, including journalling and is a very stable bit of software (which of course, is very important if you are entrusting all your data to it).
It doesn't necessarily do everything that is possible these days, though, and so the ext3 source code was forked into a new project to try and add these features, as the IBM article explains.
The biggest new thing is the maximum sizes that ext4 can support, and support for 64-bit filesystems. While the figures for the maximums in ext4 might seem astronomical at the moment (for example 1 exbibyte (EiB), or 1073741824 GiB), if trends continue, we may actually be needing to extend the current limits.
As I've already mentioned, filesystem technology has to be absolutely rock solid. It would be terrible to put all your important data into a filesystem, only to have an obscure bug corrupt it all and render it unreadable.
This is why everything has to be planned so far in advance when it comes to filesystem technology. The time it will take to get ext4 from its current experimental stage to when it is as stable as ext3.
And that's why it's a big deal. It might seem ages off yet, and many of the features will seem a bit unnecessary right now, but all this work has to happen way in advance.
